Research On Residual Fatigue Life Analysis And Prediction Of Ball Bearings

As an important category of rolling element bearing, ball bearing is widely used in mechineries. But as a vulnerable part, ball bearing is one of the primary sources of mechinery failure. The failure without precognition not only baffles the establishment of maintenance policy, but also may bring about catastrophic accidents. So the residual fatigue life prediction of ball bearing is helpful in cognizing the health of mechanics and optimizing the maintenance policy of mechanical system. Most traditional methods of residual fatigue life prediction of ball bearings are probability-based prognosis techniques which only can estimate the overall life of ball bearings. However, methods of data-driven prognosis can predict the life of a single ball bearing, but these methods are mainly dependent on state signals not faults themselves. So the accuracy and reliability of existing models are not competent when these models are used to predict the residual fatigue life of ball bearing.In accordance with these problems, supported by the National Natural Science Foundation of China and the Ministerial-Level Pre-Research Fund, based on nonlinear dynamics of ball bearings as the basis of failure analysis, improved EMD method as the mean of fault feature extraction, the fatigue life of ball bearing is analyzed. Together with microscopy failure modes of faults, combined with the contributions of fatigue life prediction theory based on Paris, this paper is devoted to research the residual fatigue life prediction of single ball bearing in the origin of failure mechanism. The main contributions of this dissertation are summarized as follows:1. Aiming at complex relationships in nonlinear dynamics of ball bearing with defects, nonlinear dynamic equations of ball bearing are established on the basis of balls as the objects and the adjoint equation which is used to describe the lean of inner ring to the outer ring is inducted. Together they can describe the dynamic characteristics of the single ball bearing perfectly. Then subsection functions and the defect impulse function are inducted and nonlinear dynamic equations of ball bearing with a single surface defect are proposed on the basis of proposed nonlinear dynamic equations of ball bearing and the impact energy equations are enduced. Then nonlinear dynamic characteristics of ball bearing with a single surface defect can be described by analytical methods and it provides a theoretical basis for investigating fatigue failure mechanism in depth.2. In order to improve the accurance of performance degenerate values of ball bearings extracted by Empirical Mode Decomposition (EMD), aiming at problems of boundary effect and modal confusion of EMD, mirror method and adding a small broadband white noise with zero mean into Intrinsic Mode Functions (IMFs) with modal confusion and then decomposed by EMD again are used to solve these problems. On this basis, combining with the improved EMD standardization process, an improved EMD method to improve the accuracy of performance degenerate values of ball bearings is proposed. Finally, the improved EMD method is applied to extract the performance degenerate values of ball bearings, together with grey model, a residual fatigue life prediction model is proposed and is verified by historial data of 6205 deep groove ball bearing. Based on this, the application of performance degenerate values in residual fatigue life prediction is analyzed and it shows that the performance degenerate values extracted by improved EMD can describe the health of ball bearing.3. According to micro-morphology of fatigue spalling, based on critical plane, the concept of critical surface is presented, then the three-dimensional crack propagation of ball bearings is transformed into two-dimensional and this enables the traditional method can be applied in them. Based on this, the fatigue life of ball bearings is divided into two parts: fatigue crack growth and fatigue spalling propagation. Then a residual fatigue life prediction model of ball bearing is proposed based on improved Paris law and its efficiency is verified by historical lives of 6205 deep groove ball bearings.4. Aiming at residual fatigue life prediction of ball bearings on line, by combining data analysis of vibration signals with investigation of micro-morphology observation and analysis of fatigue spalling, the on-line prediction of residual fatigue life of ball bearing is proposed. Aiming at the life prediction in the phase before the appearance of detectable exceptional signals, history life data is treated as a prior knowledge and introduced into the proposed prediction model as a weighting function and then a prediction model which can predict the fatigue life and residual fatigue life of ball bearing on line is proposed. Then the validity of the model is verified by experimental data.In summary, this paper launched the study on defect growth mechanism based on nonlinear dynamics theory of fault ball bearings and fault feature extraction by the improved EMD method. Based on this, the relationship between fatigue life and defect growth process is investigated. On the basis of this, utilizing the macroscopical statistics of fatigue life evaluated by traditional fatigue life prediction, combining with the advantage of Paris law which can predict the residual fatigue life of the single ball bearing, a residual fatigue life prediction model is proposed. The model can be used to predict the residual fatigue life and fatigue life of the single ball bearing on line. This fills the blank of on-line prediction methods of the single ball bearing and it is helpful for health management, failure diagnosis and condition-based maintainance.